Judith L' Reishtein, Greg Maislin, David F' Dinges, Allan I' Pack, Terri E' Weaver, and the Multisit

1
Prediction of Improvement in Sleepiness and
Function Following CPAP Treatment
Judith L. Reishtein, Greg Maislin, David F.
Dinges, Allan I. Pack, Terri E. Weaver, and the
Multisite Group University of Pennsylvania Multisi
te Group M. Mahowald, Hennepin County Medical
Center, Minneapolis MN., G. Kader , St. Lukes
Hospital, St. Louis, MO, T. Bloxham, Wichita
Clinic, Wichita, KS, C.F.P. George, Univ. Western
Ontario, Ontario, CA, H. Greenberg, Long Island
Jewish Medical Center, NY, J. Younger, Holy
Family Bon Secours Regional Health System,
Altoona ,PA
Methods
Predictor equations
Table 1 Sample Description
In the following predictor equations, variablept
is an individual patients value

• Subjects were recruited from 7 sleep clinics in
  US and Canada following diagnostic PSG if they
  were 21-60 years old, had an apnea-hypopnea index
  (AHI) gt 15 and excessive daytime sleepiness
• Following informed consent, they completed
  baseline testing Multiple Sleep Latency Test
  (MSLT), Epworth Sleepiness Scale (ESS),
  Functional Outcomes of Sleep Questionnaire
  (FOSQ), and neurobehavioral tests (not reported
  here)

1) MSLT 4.45 0.05 (AHIpt 63.9) - .44
(baseline MSLTpt 6.8) 0.10 (agept
46.7) site specific value r2 0.34, p lt
.0001 2) ESS -5.92 - 0.67 (baseline ESSpt
15.0) site specific value r2 0.44, p lt
.0001 3) FOSQ total 2.23 .02 (AHIpt
63.9) - 0.65 (baseline FOSQpt 14.7 ) site
specific value r2 0.55, p lt .0001

• Following testing session, subjects were
  instructed in use of CPAP and given machine to
  use at home
• CPAP use was monitored
• After 3 months CPAP treatment, subjects returned
  to sleep clinic for repeat of all baseline tests
• Data were analyzed using SAS.
• Variables were examined as predictors of change
  in each study outcome baseline value of outcome,
  age, BMI, AHI, site, and CPAP use. After
  calculating the means of each candidate predictor
  variable, we used these values to center each
  variable prior to inclusion in the regression
  models. As a consequence, the regression model
  intercepts could be interpreted as the predicted
  change for a person of average age, BMI, and RDI.

Conclusions
Table 2 Changes in outcomesfrom baseline to
post treatment

• CPAP use for 3 months significantly improves
  subjectively and objectively measured daytime
  sleepiness and functional status
• We demonstrate a method of predicting improvement
  with treatment from baseline condition, the
  strongest predictor for each outcome being the
  baseline value of that outcome. The site of
  testing adds to the predictive ability of the
  equations for all three outcomes, and the AHI
  adds to the predictive ability of the equations
  for the MSLT and FOSQ.
• These predictions could be used in pretreatment
  counseling to increase adherence to CPAP.
  treatment, and enhance post treatment outcome.
• Further research is needed to develop a profile
  of those who would optimally respond to
  treatment.

p lt 0.001
Funded by NHLBI, Respironics, Inc, DeVilbiss
Health Care, Nellcor Puritan Bennett,
Healthdyne Technologies, Inc